Analysis of text in job requests using natural language processing methods

Nowadays, the unabated advancement of technology also increases the developments in artificial intelligence and natural language processing. Artificial intelligence aims to improve the ability of machines to perform complex tasks such as learning, problem solving, language understanding, perception and decision making. Natural Language Processing refers to a field that focuses on the ability of computers to understand, interpret, produce and process human language. The combined use of natural language processing and artificial intelligence has enabled the emergence of innovative applications in many sectors. Within the scope of the thesis study, a study was performed to estimate the categories of job request statements in the job request application developed and used within the company. A system that makes category predictions from entered description texts has been developed. In the study, pre-trained natural language processing models Bert, DistilBERT, Electra and ConvBERT were used. Category prediction was made, after the data set was trained with these models. As a result of the study, the highest accuracy was obtained in the ConvBERT model with a success rate of 89%. In the second part of the study, the results obtained in the ConvBERT model were examined and how the system made the prediction using the Lime (Local Interpretable Model-agnostic Explanations) explainable artificial intelligence model. In this way, the prediction process in the model has been made more clear and a the system has been developed in which explainable artificial intelligence and natural language processing are used together.